Two lots of the same ester arrive. One COA says water 0.03%, the other 0.12%. Both pass a ≤0.2% limit, both go into the same emulsion. Eighteen months later one batch smells of old oil and the other does not — and the difference was on the certificate from day one.

Water content is not "volatiles"

Karl Fischer titration measures water specifically. Loss on drying measures everything that leaves the pan: residual solvent, short-chain alcohols, unreacted monomer, and water. On the same sample the two methods routinely disagree by a factor of two, so a supplier quoting "moisture ≤0.5%, loss on drying" and one quoting "water ≤0.5%, Karl Fischer" are making different promises. Ask which method before you compare numbers.

Three setups, three ranges

SetupWorking rangeSample sizeUse it for
Volumetric KF100 ppm – 100%0.1–5 gWater-rich pastes, surfactant solutions, mid-range samples
Coulometric KF1 ppm – 5%0.01–0.5 gSolvents, esters, oils, refrigerants
Oven + KFppm level1–10 gWaxes, powders, polymers that do not dissolve

The consequence is practical: a coulometric cell reports 3 ppm, a volumetric titrator cannot get near that, so writing "water ≤0.005%" against a volumetric method is a specification the lab physically cannot verify. Match the limit to the method or you are arguing about the instrument, not the product.

Where the number is actually decided

  • Sampling. A 5 g sample left open on a bench gains water while you weigh it. Take it in a sealed vial, in a dry area, and treat the vial as the sample.
  • Headspace. Moisture hides in the air above the product. Roll the drum, sample from the lower third, and never take the surface film.
  • Drift. A cell drifting above 10 µg/min is titrating the room, not the sample. Wait for a stable baseline before injecting.
  • Reagents. Ketones and some esters need methanol-free reagents; with the standard reagent they form ketals and release water, giving a number that is reproducible and wrong.
  • Solubility. If the sample does not dissolve, water inside the matrix is never released. That is what the oven method exists for.

The arithmetic that catches a wrong specification

Citric acid monohydrate (CAS 5949-29-1) carries one molecule of water per molecule of acid: MW 210.14 against 192.12 for the anhydrous form. The theoretical water is 18.02 ÷ 210.14 = 8.57%. So a purchase order reading "citric acid monohydrate, water ≤0.5%" cannot be met by any genuine monohydrate. A supplier reporting 0.4% is shipping anhydrous material, a blend, or a COA copied from the wrong grade.

The same test works on any hydrate: subtract the anhydrous molecular weight, divide by the hydrate molecular weight. It takes a calculator and saves a rejected container.

Water in an ester is a slow clock

In an ester, water hydrolyses the ester bond. The reaction is slow at 25 °C and much faster at 40 °C in a warehouse, and it produces free fatty acid: the acid value climbs while the assay falls. Every 0.1% water is roughly 1,000 ppm of potential hydrolysis, and nothing downstream removes it — you can filter a haze, you cannot un-hydrolyse an ester.

That is why low-water esters ship under nitrogen, why the headspace matters as much as the drum lining, and why a 0.10% limit on isopropyl myristate is a real request rather than a formality. It is also why water content and water activity should be read separately: a formula can have low water content and still support microbial growth. See water activity and preservation for that half of the picture.

What to write in the specification

  • Method, not just the number: "water ≤ 0.10%, Karl Fischer coulometric, ASTM E203".
  • Sample handling: "sampled from sealed drum, analysed within 24 hours".
  • Reporting limit appropriate to the method, so the lab does not report "<0.1%" for a 30 ppm result.
  • For solids and waxes, state the oven temperature — the same sample gives different results at 105 °C and 130 °C.

We supply low-water isopropyl myristate (CAS 110-27-0), caprylic/capric triglyceride (CAS 65381-09-1), polysorbate 80 (CAS 9005-65-6), PEG-40 hydrogenated castor oil (CAS 61788-85-0), cetearyl alcohol (CAS 67762-27-0), glyceryl caprylate (CAS 26402-26-6) and white mineral oil 22# (CAS 8012-95-1), plus citric acid anhydrous (CAS 77-92-9), citric acid monohydrate (CAS 5949-29-1) and calcium lactate (CAS 814-80-2). Water is reported by method on every COA; tell us the limit and the method and we will confirm against the batch.